sign in sign up
<<
Home , Cyclin, Cyclin A2, Cyclin B1, Cytokinesis, G1 phase, G2 phase

-dependent 1 (Cdk1) is essential for division and suppression of DNA re-replication but not for regeneration

M. Kasim Dirila, Chandrahas Koumar Ratnacarama, V. C. Padmakumarb, Tiehua Duc, Martin Wasserc,d, Vincenzo Coppolab,1, Lino Tessarollob, and Philipp Kaldisa,e,2

aInstitute of Molecular and Cell , Agency for Science, Technology, and Research (A*STAR), Republic of Singapore 138673; Departments of eBiochemistry and dBiological Sciences, National University of Singapore, Republic of Singapore 117597; bMouse Program, National Cancer Institute, Frederick, MD 21702-1201; and cBioinformatics Institute, Agency for Science, Technology, and Research (A*STAR), Republic of Singapore 138671

Edited by Charles J. Sherr, Howard Hughes Medical Institute and St. Jude Children’s Research Hospital, Memphis, TN, and approved January 13, 2012 (received for review September 16, 2011)

Cyclin-dependent kinase 1 (Cdk1) is an archetypical kinase and viable homozygous knockout mice or at or E10.5, a central regulator that drives cells through and . indicating that Cdk1 is an essential for early embryonic Knockouts of Cdk2, Cdk3, Cdk4, or Cdk6 have resulted in viable development. However, genotyping of at E3.5 mice, but the in vivo functions of Cdk1 have not been fully ex- revealed that 16% were homozygous knockout, 56% heterozy- gous, and 28% wild type for the Cdk1 locus (Fig. 1B). Detailed plored in . Here we have generated a conditional-knock- NULL out mouse model to study the functions of Cdk1 in vivo. Ablation analysis of Cdk1 blastocysts by microscopy revealed that they had a reduced number of nuclei, which were larger in size of Cdk1 leads to arrest of around the fi compared with wild-type and heterozygous littermates (Fig. 1 C stage. Interestingly, liver-speci c deletion of Cdk1 is and D). The first few cells divisions in Cdk1-deficient embryos well tolerated, and liver regeneration after partial hepatectomy are possibly achieved because of the presence of maternally is not impaired, indicating that regeneration can be driven by cell deposited Cdk1 mRNA in the (18). After the maternally growth without . The loss of Cdk1 does not affect S provided mRNA runs out, cells of the Cdk1NULL embryos were phase progression but results in DNA re-replication because of an not able to divide anymore, which suggests Cdk1 is essential for increase in Cdk2/ activity. Unlike other Cdks, loss of Cdk1 cell division in early-stage embryos. in the liver confers complete resistance against tumorigenesis in- To confirm that Cdk1 is essential for , we duced by activated Ras and silencing of . turned to primary mouse embryonic fibroblasts (MEFs), in which we could induce the loss of Cdk1 by addition of 4-hydrox- – cancer | knockout mice | regulation | ytamoxifen (4-OHT) (Fig. S1 C H). When these MEFs were analyzed for proliferation by using techniques such as ala- marBlue or 3T3 assays, Cdk1NULL MEFs did not increase in he activities of cyclin-dependent (Cdks) control all numbers but entered senescence prematurely, demonstrating Taspects of cell division, including entry into the cell cycle from that Cdk1 is essential for cell proliferation (Fig. 1 E and F). quiescence, the G1/ transition, DNA replication in S phase, nuclear breakdown, condensation and seg- Liver Regeneration in the Absence of Cell Division. Because of the regation, and (1). The mammalian contains embryonic lethality, we generated mice that were specifically at least 20 different Cdks, and widespread compensation among targeted to ablate the expression of Cdk1 in the liver (hep- Cdks and has been reported (2, 3). Cdk1 was the first Cdk atocytes; Cdk1FLOX/FLOX mice expressing the albumin-Cre − − identified (4, 5), is conserved in all , plays important transgene, referred to as Cdk1Liv / ). We chose the liver as an roles in mitosis, and can drive S phase in the absence of Cdk2 in vivo model system because hepatocytes rarely divide in adult (6). Mouse knockouts of Cdk2 (7, 8), Cdk3 (9), Cdk4 (10, 11), or unless regeneration is induced by partial hepatectomy Cdk6 (12, 13) are viable, indicating that any single Cdk in- (PH; ref. 19 and below). Albumin-Cre expression starts during dependently is not essential for survival of mice. Interestingly, embryonic development but only reaches sufficiently high levels double knockouts of Cdk2/Cdk4 (14) and Cdk4/Cdk6 (13) but to cause efficient recombination approximately 3 wk after birth not Cdk2/Cdk6 (13) are embryonic lethal, suggesting genetic (20), a time when liver development and hematopoiesis are al- − − interactions. Therefore, Cdk2 and Cdk4 (as well as Cdk4 and ready completed. Cdk1Liv / mice were viable and did not dis- Cdk6) display overlapping functions and can substitute for each play any adverse phenotypes. We prepared Feulgen-stained − − other (3, 15). In the absence of Cdk2 and Cdk4, we have dem- sections of Cdk1Liv / and control (Fig. 2A). The diameter − − onstrated an accumulation of hypophosphorylated Retinoblas- of the Cdk1Liv / nuclei was increased by 28%, and the cell density was decreased by 35%. To force hepatocytes to reenter toma (Rb), which represses -mediated − − the cell cycle, we subjected Cdk1Liv / mice to 70% PH. In- leading to decreased expression of E2F-target like Cdk1 − − and others (14). terestingly, PH in both wild-type and Cdk1Liv / mice led to Constitutive gene disruption in the Cdk1 locus have complete regeneration of liver mass within 3 wk (Fig. 2 B and C). been reported, but neither of these approaches yielded homo- However, Cdk1-deficient hepatocytes were unable to enter zygous Cdk1NULL cells or tissues for further analysis (16, 17). To study the in vivo functions of Cdk1, tissue specificity, and re- quirement in adult mice, we have generated conditional knock- Author contributions: P.K. designed research; M.K.D., C.K.R., and V.C.P. performed re- out mice by inserting LoxP sites on both sides of exon 3 in the search; V.C. and L.T. contributed new reagents/analytic tools; T.D. and M.W. analyzed Cdk1 locus (Fig. 1A and Fig. S1 A and B). data; and P.K. wrote the paper. The authors declare no conflict of interest. Results This article is a PNAS Direct Submission. Cdk1 Is Essential for Cellular Proliferation and Early Embryonic 1Present address: Department of Molecular Virology, , and Medical Genetics, Development. To investigate the effects of loss of Cdk1 in Ohio State University, Columbus, OH 43210. +/NULL mouse development, we crossed Cdk1 animals with each 2To whom correspondence should be addressed. E-mail: [email protected]. other and monitored the progeny at postnatal day 21 (P21), This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. embryonic day 10.5 (E10.5), and E3.5. We could not detect 1073/pnas.1115201109/-/DCSupplemental.

3826–3831 | PNAS | March 6, 2012 | vol. 109 | no. 10 www.pnas.org/cgi/doi/10.1073/pnas.1115201109 Downloaded by guest on September 27, 2021 mitosis (Fig. S2), indicating that liver regeneration can be ach- ieved by in the absence of cell division. PH induced − − expression of Cdk1 in control but not in Cdk1Liv / animals (Fig. 2D; note that the Cdk1 band in lanes 7–12 originates from liver cells other than hepatocytes). In control mice, the cell density and nuclear diameter did not change significantly by 96 h after PH (Fig. 2 E and F). In contrast, we measured a 27% decrease in cell density (Fig. 2 E and G) and a 67% increase in nuclear di- ameter (Fig. 2 F and G), which is consistent with ablated cell − − division and a substantial increase in cell growth in Cdk1Liv / mice. As a measure of DNA replication after PH, we deter- mined BrdU incorporation and detected a peak at 48 h after PH (Fig. 2 H and I), which is consistent with published results (21). − − Cdk1Liv / livers displayed BrdU incorporation rates similar to − − those of wild type, and therefore we hypothesized that Cdk1Liv / cells must be able to undergo some form of DNA replication.

Cdk1 Is Dispensable for DNA Replication but Regulates Cdk2 Activity to Suppress DNA Re-replication. DNA replication in S phase is promoted by Cdk2 in complex with cyclins E and A2 (22). However, there are no detectable replicative deficiencies in Cdk2- knockout mice because Cdk1 can fully compensate for the ab- sence of Cdk2 (6–8). To further investigate whether Cdk1 activity is required for initiation and progression of DNA replication in presence of Cdk2, we turned to our conditional knockout primary MEFs, in which we can study cell-cycle progression kinetics in a controlled fashion (Fig. S1 C and D). MEFs were synchronized in the G0/ by serum star- vation, and Cdk1 knockout was simultaneously induced by addi- tion of 4-OHT. After synchronization, MEFs were released into S phase in the presence of serum and labeled by BrdU as a measure for DNA replication. During the first cell cycle, Cdk1NULL cells progressed through S phase like wild-type MEFs did (Fig. 3 A and B and Fig. S3). Therefore, the loss of Cdk1 does not affect S phase progression in the presence of wild-type Cdk2. However, Cdk1NULL MEFs cannot initiate early events of mitotic entry such as cytoskeletal reorganization and rounding up of the cell body (Fig. S1H) and therefore arrest in G2 without entering mitosis. A detailed analysis of the FACS data indicated an additional 8N peak, suggesting DNA re-replication in the absence of cell division (Fig. 3C). We found that ∼30% of the Cdk1-deficient cells had undergone DNA re-replication, compared with 6% in wild-type MEFs (Fig. 3C). Simultaneous knockdown of Cdk2 in Cdk1NULL cells significantly reduced re-replication (Fig. 3D), suggesting that Cdk2 substantially contributed to DNA re-replication. To investigate the underlying mechanisms of re-replication, we measured the kinase activity after immunoprecipitating Cdks or FLOX Fig. 1. Generation and analysis of Cdk1 conditional knockout mice and cyclins. In Cdk1 cells (these are essentially wild-type cells), cells. (A) The Cdk1 genomic locus (I) was modified in ES cells with the tar- all activities peaked in the 24- to 48-h range (Fig. 4A, lanes 1–6; geting vector (II) shown. An FRT-flanked neomycin-selection cassette was for quantitative analysis of kinase activities, see Fig. S4). In introduced along with LoxP recombination sites (red triangles) on both sides contrast, there was no detectable Cdk1 or -associated NULL of exon 3, which generated a mutant Cdk1 locus (III). For Southern blot kinase activity at any of the time points in Cdk1 cells (Fig. analysis, 5′ and 3′ probes located outside of the targeting vector were used, 4A, top two gels, lanes 7–12). This finding suggests that cyclin B1 after an EcoRV (RV) digest, resulting in a 31,206-bp (recombinant) and does not form active complexes with Cdk2 despite its increased a 9,110-bp (5′) or 20,300-bp (3′) fragment for wild type (Fig. S1A). Upon association in Cdk1NULL cells (Fig. 4C and Fig. S4B). Inter- expression of FLP recombinase, the neomycin cassette was removed, and estingly, Cdk2 or cyclin A2-associated kinase activities were el- only the LoxP sites remained in the locus (IV, Cdk1FLOX). The levels of Cdk1 evated compared with control and extended from 18 to 72 h (Fig. protein expression were similar in Cdk1FLOX and Cdk1WT mice (Fig. S1B). 4A, third and fourth gels from the top, lanes 8–12). This result After Cre recombinase expression, exon 3 was excised (V), which resulted in indicates that Cdk1 might be responsible for suppressing Cdk2/ deletion of Cdk1 and a frame shift. PCR genotyping primers are indicated cyclin A2 kinase activity at the end of S phase in wild-type cells, (Pr1, -2, and -3), and the sequences can be found in Table S1.(B) To generate fi FLOX which prevents re-replication from occurring and was con rmed Cdk1 knockouts, Cdk1 mice were crossed with β-–Cre mice ex- by silencing of Cdk2 (Fig. 3D).

pressing Cre recombinase ubiquitously in all tissues, including germ line. The WT/NULL The increased Cdk2 activity in the absence of Cdk1 could be resulting Cdk1 mice were interbred, and the offspring was analyzed caused by several factors, including decreased binding of Cdk at weaning (P21), midgestation (E10.5), or blastocyst (E3.5) stage. (C and D) NULL/NULL inhibitors, decreased inhibitory , increased Cdk2 Cdk1 blastocysts were visualized by Hoechst of their nuclei or cyclin A2 expression (increased transcription or decreased followed by fluorescence microscopy. Comparison of Cdk1-deficient blasto- cysts with heterozygous or wild-type littermates indicated a reduced number degradation), increased Cdk complex formation, or differential of cells (C); however, their nuclei were larger in size (D). (E and F) Three independent MEF lines (Fig. S1 C–H) were treated with 4-OHT to induce Cdk1 knockout, and their proliferative potential was monitored by 3T3 and ala- premature onset of that was detected by senescence- marBlue proliferation assays for several passages or days, respectively (E). associated β-gal staining (F). Therefore, cells lacking Cdk1 cannot proliferate Deletion of Cdk1 resulted in a rapid arrest of cellular proliferation and but instead enter a senescent state and survive in culture medium.

Diril et al. PNAS | March 6, 2012 | vol. 109 | no. 10 | 3827 Downloaded by guest on September 27, 2021 Fig. 2. Liver-specific knockout of Cdk1. (A) Liver sections from Cdk1Liv−/− and control mice were stained with Feulgen to visualize their nuclei. Even though the overall liver size was similar, Cdk1-knockout livers contained fewer hep- atocytes with enlarged nuclei. (B and C) To analyze the regenerative potential and S phase entry in Cdk1-knockout livers, 70% of the liver mass was removed by PH. Animals were euthanized 4 or 21 d later, and the ratio of liver to body weight was determined. Each dot in the chart rep- resents an individual PH experiment, and mean values are − − depicted by the black or red lines. Cdk1Liv / livers regen- erated the lost mass within 3 wk, comparable to controls. (D)Toconfirm Cdk1 knockout, Western blots with Cdk1 were performed at different times after PH in control and Cdk1Liv−/− liver extracts. Cdk1 expression detected in knockout livers is due to liver cells other than − − hepatocytes. (E–G) Cdk1Liv / display a decreased hepato- cyte density compared with controls and the difference is − − exacerbated by 96 h after PH (E). Cdk1Liv / hepatocytes have enlarged nuclei that become even bigger after PH (F and G). (H and I) Regenerating livers were pulse-labeled with BrdU at 48 (H), 72, and 96 h after PH. BrdU in- corporation was detected by immunohistochemical stain- ing of liver sections (H) and quantified with a custom developed image analysis software (I). Data were obtained from n = 3 animals in E, F, and I and are represented as mean ± SD. Black and red bars indicate Cdk1FLOX and Cdk1Liv−/− livers, respectively.

activity of Cdk2/cyclin A2 and Cdk1/cyclin A2 complexes. We specific kinase activity of Cdk1/cyclin A2 is different from Cdk2/ tested all these possibilities and conclude that the latter two cyclin A2. Therefore, we determined the kinetic parameters of possibilities contributed significantly to the increase of Cdk2/ these Cdk complexes with purified and found that the cyclin A2 activity in the absence of Cdk1 (as described below). Km ( H1) for Cdk1/cyclin A2 is about two times lower and The mRNA levels of Cdk2, cyclin A2, and many other cell-cycle– the Vmax () is approximately eight times lower than related genes were comparable between Cdk1NULL and Cdk2/cyclin A2 (Fig. S8 A and B). The activity of Cdk1/cyclin A2 Cdk1FLOX cells. As expected, the Cdk1 mRNA was almost un- complexes is very low compared with Cdk2/cyclin A2, thus ef- detectable after deletion of Cdk1 (Fig. S7). The protein levels of fectively resulting in suppression of the Cdk2 activity at the end Cdk2 and cyclin A2 were similar in the absence of Cdk1 com- of S phase to prevent re-replication. This finding was corrobo- pared with control cells (Fig. 4B), and therefore changes in rated by experiments in Cdk2NULL MEFs, in which the cyclin A2- protein synthesis or degradation were not causing the increase in associated kinase activity resulting exclusively from Cdk1/cyclin Cdk2/cyclin A2 activity. In addition, the levels of p21 and p27 A2 complexes was substantially lower compared with wild-type were marginally increased in the absence of Cdk1 (Fig. 4B). We cells (Fig. S6). The re-replication phenotype that we uncovered also analyzed the inhibitory phosphorylation (Y15PO ) on Cdk2 in MEFs is most likely the reason for the increased polyploidy 4 − − and observed an increase of tyrosine 15 phosphorylation per after PH in Cdk1Liv / livers (Fig. 2). molecule of Cdk2 in the absence of Cdk1 (Fig. 4C). Analysis of Cdk inhibitor binding to either Cdk2 (Fig. 4C) or cyclin A2 (Fig. Tumorigenesis Requires Cdk1 Activity. Analysis of tumor 4D) did not reveal a difference for p27. However, we observed samples revealed that expression of several Cdks and cyclins are a slightly increased binding of p21 in the Cdk1NULL cells com- often up-regulated (23). However, gene-knockout studies by us pared with control cells, which would have an inhibitory effect on and others (2, 15, 24) have indicated that none of the previously the kinase activity. Finally, we determined the number of Cdk2/ tested Cdks can completely prevent tumorigenesis. Therefore, cyclin A2 complexes [either immunoprecipitated Cdk2/Western- we determined whether the loss of Cdk1 prevents cellular blotted cyclin A2 (Fig. 4C) or immunoprecipitated cyclin A2/ transformation and tumor development in vivo. First, we tested Western-blotted Cdk2 (Fig. 4D)] and detected a substantial in- the effect of loss of Cdk1 on cell transformation. Expression of crease in the absence of Cdk1. The increased number of Cdk2/ activated Ras and dominant-negative p53 or shRNA against p53 cyclin A2 complexes is consistent with the increased kinase ac- resulted in substantial increases in cell proliferation and trans- tivity observed (Fig. 4A). Because more cyclin A2 molecules are formed colonies in control MEFs (Fig. 5A). In cells treated with binding to Cdk2 in the absence of Cdk1, most likely the majority 4-OHT to induce the loss of Cdk1, few colonies were observed. of the cyclin A2 molecules are bound to Cdk1 in wild-type cells. In addition, the few colonies that were detected expressed Cdk1 Thus, it appears that, as cells progress through S phase, cyclin A2 because of lack of full recombination. Therefore, we concluded initially forms complexes with Cdk2, and, later, most cyclin A2 that cell proliferation in transformed cells depends on Cdk1. molecules are bound to Cdk1 (Fig. 3E). Cdk2/cyclin A2 activity To investigate tumor formation in vivo, we used tail-vein − needs to be inhibited after S phase to prevent re-replication (Fig. injections of activated Ras alone (in a p53+/ background) or in 3E). Thus, we hypothesize that, to achieve this inhibition, the combination with an shRNA to silence p53 expression (25).

3828 | www.pnas.org/cgi/doi/10.1073/pnas.1115201109 Diril et al. Downloaded by guest on September 27, 2021 Fig. 3. Cdk1 is redundant for S phase progression, but its loss results in . (A) Cdk1FLOX MEFs were synchronized in the G0/G1 phase by serum starvation for 72 h and simultaneously treated with 4-OHT to induce knockout of Cdk1. They were then released into S phase by addition of serum, pulse- labeled with BrdU, and analyzed by FACS to determine the percentage of cells in S, G1, and G2 phases. Histograms depict the distribution of DNA content as detected by propidium iodide (PI) staining. Insets show the distribution of BrdU incorporation (BrdU) versus DNA content (PI) within the same population. Three independent experiments were performed with three different clones. (B) Quantitative analysis of cells in S phases of the cell cycle are shown. Although Cdk1-knockout MEFs seem to have a slightly reduced ratio in S phase at 18 and 24 h, this is due to the accumulation of G2-arrested cells during the serum starvation period, which cannot enter S phase as efficiently as G1 cells. Data are represented as mean ± SD. (C) Deletion of Cdk1 resulted in endoreduplication and accumulation of cells with a DNA content greater than 4N. (D) However, simultaneous removal of Cdk2 with shRNAs partially rescued the endor- eduplication phenotype. See also Fig. S3.(E) A model demonstrating how Cdk1 manages to reduce Cdk2 activity at the end of S phase and prevents DNA re-replication. According to our model, kinase activity of Cdk2/cyclin A2 complexes is higher than that of Cdk1/cyclin A2 complexes. Cyclin A2-associated kinase activity peaks in S phase. However, increased levels of Cdk1 protein during late S to early G2 phase results in fewer cyclin A2 molecules available for binding to Cdk2. As a result, Cdk1 quenches cyclin A2-associated kinase activity. In Cdk1-deficient cells (represented by dashed lines), cyclin A2 molecules are forming complexes only with Cdk2. Therefore, Cdk2/cyclin A2 kinase activity cannot be suppressed and persists throughout G2 phase.

Control mice (Cdk1FLOX/FLOX) developed many liver tumors Discussion within 3–4 mo (Fig. 5 B, D, and E). In contrast, none of the − − Previous work has suggested that Cdk2, Cdk4, or Cdk6 are not Cdk1Liv / (Cdk1FLOX/FLOX; albumin-Cre) mice developed essential to complete a full cell division cycle. Cdk1 can com- tumors, even after aging them up to 9 mo (Fig. 5 C, D, and E). pensate for their combined loss by forming active complexes with Only one of the Cdk1 liver-knockout animals developed liver A-, B-, E-, and D-type cyclins (6, 24, 27). Because constitutive tumors, but further analysis by genotyping of the tumor samples inactivation of the Cdk1 locus led to early embryonic lethality, revealed incomplete Cdk1 knockout. Histopathological analysis we had to resolve to generate a conditional knockout mouse − − indicated that Cdk1Liv / livers appeared similar to livers from model for detailed analysis of loss of Cdk1. untreated mice, whereas massive tumors and frequent neoplasia Here we demonstrate that, although Cdk1 is essential for cell were detected in control mice (Fig. 5E). proliferation and tumorigenesis, it is not required for DNA To analyze tumor cells in vitro, we isolated them from Cdk1FLOX/ replication and liver regeneration. Earlier studies wherein Cdk1 FLOX liver tumors. These primary liver tumor cells grew well (Fig. activity was targeted by different methods such as knock-in of S9A) and were able to form colonies (Fig. 5F, Left). We infected a temperature-sensitive in human cells (28), selective drug

these cells with a retrovirus encoding CreERT2 and treated them inhibitors (26), or RNAi-mediated silencing (29) have shown that CELL BIOLOGY with 4-OHT to induce the loss of Cdk1 (Fig. S9B). After 4-OHT loss or inhibition of Cdk1 activity results in DNA re-replication, treatment, these Cdk1NULL cells were not able to form any colonies polyploidy, and cellular senescence but only in transformed cell (Fig. 5F, Right). Identical experiments were performed with the lines in vitro. We extended these findings in vivo and uncovered Cdk1-specific inhibitor RO-3306 (26) and roscovitine (a general a molecular mechanism by which Cdk1 suppresses re-replication Cdk inhibitor) instead of 4-OHT, and comparable results were through inhibition of Cdk2 activity by sequestering cyclin A2 obtained (Fig. S9C). Cdk1-deficient tumor cells rapidly became molecules into low-activity complexes (Fig. 3E). This finding is senescent (Fig. S9D), as we had previously observed for MEFs (Fig. valid not only for MEFs but also in liver tumor cells (Fig. S5), and 1F). Our results indicate that Cdk1 is required for the growth of therefore represents an additional layer of regulation of Cdk ac- liver tumors in vitro and in vivo. tivities during the cell cycle.

Diril et al. PNAS | March 6, 2012 | vol. 109 | no. 10 | 3829 Downloaded by guest on September 27, 2021 Fig. 4. Increased Cdk2 kinase activity in Cdk1-deficient MEFs. (A) To analyze the kinase activities associated with Cdks and cyclins, protein extracts were prepared from cells in Fig. 3 at different time points and subjected to immu- noprecipitation with the indicated antibodies followed by in vitro kinase assays using radiolabeled ATP and histone H1 as substrates. Cdk2 and cyclin A2-associated kinase activities were significantly increased in Cdk1-deficient cells. (B) Samples from A were subjected to SDS/PAGE and Western blotting with the indicated antibodies. (C and D) To analyze the Cdk2/cyclinA2 complexes, protein extracts were immu- noprecipitated with antibodies against Cdk2 (C) or cyclin A2 (D) followed by Western blotting with the indicated anti- bodies. See also Figs. S4–S6 and S8.

Cdk1 and Re-replication. Cdks have pleiotropic roles in initiation of Cdk1 activity during mitosis is essential for preventing the re- S phase entry and prevention of re-replication after DNA syn- replication. Inhibition of Cdk1 activity via mutagenesis (28, 31) thesis has been completed. Although Cdk-independent mecha- or selective drug inhibitors (26) has been shown to result in in- nisms exist to avoid refiring of replication origins (30), coupling creased endoreduplication. Corroborating this hypothesis, p57- of the whole process with the cell cycle requires sequential ac- mediated inhibition of Cdk1 activity in an endogenous system, tivation of Cdk2 and Cdk1 to initiate origin firing and ensuring trophoblast giant cells, has been shown to be essential for re- that it takes place only once per cycle. peated rounds of endoreduplication that take place during their

Fig. 5. Oncogenic transformation or tumorigenesis is not possible in the absence of Cdk1. (A) Cdk1FLOX MEFs were immortalized or transformed with p53 shRNA or activated Ras/p53DN, respectively. Colony-formation assays after induction of the cells with 4-OHT indicate that Cdk1-deficient cells cannot be − − immortalized or transformed. (B–E) Liver tumors were induced by tail-vein injection of activated Ras (B), but no tumors were detected in Cdk1Liv / livers (C). Quantitative analysis of liver tumors in mice of the indicated genotypes (D) is shown. Histological sections of the liver were stained with hematoxylin and − − eosin (E) to demonstrate normal morphology in the Cdk1Liv / (Right) and many tumors in control livers (Left). (F) Tumors from two different Cdk1FLOX/FLOX mice (LI37 and LI53) were isolated, dissociated, and grown in culture. After introduction of CreERT2 and treatment with DMSO (control) or 4-OHT, cells were subjected to a colony-formation assay followed by Giemsa staining.

3830 | www.pnas.org/cgi/doi/10.1073/pnas.1115201109 Diril et al. Downloaded by guest on September 27, 2021 maturation (32); in the same system, the authors have also shown Isolation and Culture of Primary MEFs. Primary MEFs were isolated from E13.5 that Cdk2 is required for endoreduplication. mouse embryos as described previously (7). Briefly, the head and the visceral In our current work, we demonstrate that Cdk1-deficient pri- organs were removed, the embryonic tissue was chopped into fine pieces mary MEFs and hepatocytes are more likely to undergo endor- with a razor blade and trypsinized for 15 min at 37 8C, and finally tissue and eduplication compared with controls. Simultaneous inactivation cell clumps were dissociated by pipetting. Cells were plated in a 10-cm cul- of Cdk2 by RNAi in Cdk1NULL MEFs reverted the re-replication ture dish (passage 0) and grown in DMEM (12701-017; Invitrogen) supple- phenotype. We propose a mechanism to explain why Cdk1-de- mented with 10% FCS (26140; Invitrogen) and 1% penicillin/streptomycin ficient cells endoreduplicate their and how Cdk1 sup- (15140-122; Invitrogen). Primary MEFs were cultured in a humidified in- presses Cdk2 activity at the end of S phase to aid prevention of cubator with 5% CO2 and 3% O2. replication-origin refiring (Fig. 3E). Tail-Vein Injections and Liver Tumorigenesis. Hydrodynamic tail-vein injections Cdk1 and Tumorigenesis. We have demonstrated that MEFs cannot and Sleeping Beauty transposon-induced liver tumorigenesis were performed be transformed after deletion of Cdk1. Validation of these results as described previously (25, 35). Six- to 10-wk-old mice were injected with in vivo proved that liver tumors do not form in absence of Cdk1. a plasmid mixture diluted in lactated Ringer’s solution. Animals were injec- Because Cdk1 seems to be essential for proliferation in each and ted with a volume corresponding to 10% of their body weight (i.e., 2 mL every cell and tissue type tested, its inactivation would prevent for a 20-g mouse, not exceeding 2.5 mL) through their lateral tail vein within formation and propagation of tumors. This finding is in contrast to 8–10 s using 27-gauge needles. Plasmids encoding 12.5 μg of transposase other Cdks, whose inactivation had little effect on tumor forma- (pGK-SleepingBeauty13) and a total of 25 μg of transposon (pT2-Caggs- tion (6, 24, 33). Therefore, our results indicate the potential of NRasV12 and pT2-sh p53) were injected per . Plasmid DNA used for Cdk1 inhibitors in cancer therapy if we can prevent the detri- injection was purified with the Qiagen EndoFree Plasmid Maxi Kit mental side effects resulting from unintentionally interfering with (12362). Control mice (those expressing Cdk1 in their livers) were euthanized the essential functions of Cdk1 in proliferative tissues. when moribund (usually within 3–4 mo) or kept for a maximum of 6 mo. Test animals (those that are Cdk1 knockout in their livers) were kept for a maxi- Materials and Methods mum of 9 mo before euthanizing for histological analysis. All animal exper- iments were done in compliance with the Institutional Animal Care and Use SI Materials and Methods include details on primers, genotyping, real-time Committee guidelines. PCR, Cdk1 conditional knockout mice and other transgenic lines, blastocyst isolation, Western blot analysis, immunoprecipitation, kinase assays, , FACS analysis, PH, and image analysis. ACKNOWLEDGMENTS. We thank Eileen Southon and Susan Reid for help in generating the Cdk1FLOX/FLOX mice, and David Largaespada for transposase/ transposon constructs. We appreciate that Jos Jonkers and Anton Berns pro- Generation of Cdk1 Conditional Knockout Mice. Mouse genomic DNA har- vided the ROSA26-CreERT2 mice, Andy McMahon the Cre-Esr1 mice, Mark boring the Cdk1 locus was isolated from the BAC clone 305J21 (ResGen). Using Lewandoski the β-actin–Cre/Flpe mice, and T. Jake Liang the albumin-Cre recombineering technique (34), LoxP recombination sites and a neomycin- mice. We thank Nancy Jenkins and Neal Copeland for advice, suggestions, selection cassette were introduced flanking the third coding exon of the reagents, and support. We are thankful to Cyril Berthet for reagents and mouse Cdk1 genomic locus. The resulting targeting vector was linearized by discussion as well as to Steve Cohen and Neal Copeland for comments on the NotI digestion, and ES cells were electroporated. After positive and negative manuscript. We also thank Davor Solter and Barbara Knowles for technical advice; June Wang, Chloe Sim, and Vithya Anantaraja for animal care; Keith selection with Geneticin and ganciclovir, respectively, genomic DNA of sur- Rogers, Susan Rogers, and the technicians of the /Histotechnology viving ES cell colonies were screened for by Laboratory for superb analysis of mouse pathology; and the P.K. laboratory Southern hybridization (Fig. S1A). Correctly targeted ES cell clones were for support and discussions. This work was supported by the Biomedical identified and used for the generation of the Cdk1 conditional knockout Research Council of the Agency for Science, Technology, and Research mouse strain. (A*STAR), Singapore.

1. Morgan DO (2007) The Cell Cycle: Principles of Control (New Science Press, London). 19. Fausto N, Campbell JS (2003) The role of hepatocytes and oval cells in liver re- 2. Malumbres M, Barbacid M (2009) Cell cycle, CDKs and cancer: A changing paradigm. generation and repopulation. Mech Dev 120:117–130. Nat Rev Cancer 9:153–166. 20. Postic C, Magnuson MA (2000) DNA excision in liver by an albumin-Cre transgene 3. Satyanarayana A, Kaldis P (2009) Mammalian cell-cycle regulation: Several Cdks, nu- occurs progressively with age. Genesis 26:149–150. merous cyclins and diverse compensatory mechanisms. 28:2925–2939. 21. Satyanarayana A, Hilton MB, Kaldis P (2008) p21 Inhibits Cdk1 in the absence of Cdk2 4. Lohka MJ, Hayes MK, Maller JL (1988) Purification of maturation-promoting factor, an to maintain the G1/S phase DNA damage checkpoint. Mol Biol Cell 19:65–77. intracellular regulator of early mitotic events. Proc Natl Acad Sci USA 85:3009–3013. 22. Labib K (2010) How do Cdc7 and cyclin-dependent kinases trigger the initiation of 5. Nurse P, Thuriaux P (1980) Regulatory genes controlling mitosis in the fission chromosome replication in eukaryotic cells? Genes Dev 24:1208–1219. – Schizosaccharomyces pombe. Genetics 96:627–637. 23. Malumbres M, Barbacid M (2007) Cell cycle kinases in cancer. Curr Opin Genet Dev 17:60 65. 6. Aleem E, Kiyokawa H, Kaldis P (2005) Cdc2- complexes regulate the G1/S phase 24. Padmakumar VC, Aleem E, Berthet C, Hilton MB, Kaldis P (2009) Cdk2 and Cdk4 activities – transition. Nat Cell Biol 7:831–836. are dispensable for tumorigenesis caused by the loss of p53. Mol Cell Biol 29:2582 2593. 7. Berthet C, Aleem E, Coppola V, Tessarollo L, Kaldis P (2003) Cdk2 knockout mice are 25. Carlson CM, Frandsen JL, Kirchhof N, McIvor RS, Largaespada DA (2005) Somatic in- viable. Curr Biol 13:1775–1785. tegration of an oncogene-harboring Sleeping Beauty transposon models liver tumor – 8. Ortega S, et al. (2003) Cyclin-dependent kinase 2 is essential for but not for development in the mouse. Proc Natl Acad Sci USA 102:17059 17064. 26. Vassilev LT, et al. (2006) Selective small-molecule inhibitor reveals critical mitotic mitotic cell division in mice. Nat Genet 35:25–31. functions of human CDK1. Proc Natl Acad Sci USA 103:10660–10665. 9. Ye X, Zhu C, Harper JW (2001) A premature-termination in the Mus mus- 27. Bienvenu F, et al. (2010) Transcriptional role of in development revealed by culus cyclin-dependent kinase 3 gene. Proc Natl Acad Sci USA 98:1682–1686. a genetic-proteomic screen. Nature 463:374–378. 10. Rane SG, et al. (1999) Loss of Cdk4 expression causes insulin-deficient diabetes and 28. Itzhaki JE, Gilbert CS, Porter ACG (1997) Construction by gene targeting in human Cdk4 activation results in β-islet cell hyperplasia. Nat Genet 22:44–52. cells of a “conditional” CDC2 mutant that rereplicates its DNA. Nat Genet 15:258–265. 11. Tsutsui T, et al. (1999) Targeted disruption of CDK4 delays cell cycle entry with en- 29. L’Italien L, Tanudji M, Russell L, Schebye XM (2006) Unmasking the redundancy be- hanced p27Kip1 activity. Mol Cell Biol 19:7011–7019. tween Cdk1 and Cdk2 at G2 phase in human cancer cell lines. Cell Cycle 5:984–993. 12. Hu MG, et al. (2009) A requirement for cyclin-dependent kinase 6 in thymocyte de- 30. Arias EE, Walter JC (2007) Strength in numbers: Preventing rereplication via multiple – velopment and tumorigenesis. Cancer Res 69:810 818. mechanisms in eukaryotic cells. Genes Dev 21:497–518. 13. Malumbres M, et al. (2004) Mammalian cells cycle without the D-type cyclin-de- 31. Broek D, Bartlett R, Crawford K, Nurse P (1991) Involvement of p34cdc2 in establishing – CELL BIOLOGY pendent kinases Cdk4 and Cdk6. Cell 118:493 504. the dependency of S phase on mitosis. Nature 349:388–393. 14. Berthet C, et al. (2006) Combined loss of Cdk2 and Cdk4 results in embryonic lethality 32. Ullah Z, Kohn MJ, Yagi R, Vassilev LT, DePamphilis ML (2008) Differentiation of tro- – and Rb hypophosphorylation. Dev Cell 10:563 573. phoblast stem cells into giant cells is triggered by p57Kip2 inhibition of CDK1 activity. fi 15. Berthet C, Kaldis P (2007) Cell-speci c responses to loss of cyclin-dependent kinases. Genes Dev 22:3024–3036. Oncogene 26:4469–4477. 33. Martín A, et al. (2005) Cdk2 is dispensable for cell cycle inhibition and tumor sup- 16. Santamaría D, et al. (2007) Cdk1 is sufficient to drive the mammalian cell cycle. Nature pression mediated by p27Kip1 and p21Cip1. Cancer Cell 7:591–598. 448:811–815. 34. Lee E-C, et al. (2001) A highly efficient Escherichia coli-based chromosome engi- 17. Satyanarayana A, et al. (2008) Genetic substitution of Cdk1 by Cdk2 leads to em- neering system adapted for recombinogenic targeting and subcloning of BAC DNA. bryonic lethality and loss of meiotic of Cdk2. Development 135:3389–3400. Genomics 73:56–65. 18. Evsikov AV, et al. (2006) Cracking the egg: Molecular dynamics and evolutionary aspects 35. Bell JB, et al. (2007) Preferential delivery of the Sleeping Beauty transposon system to of the transition from the fully grown to . Genes Dev 20:2713–2727. livers of mice by hydrodynamic injection. Nat Protoc 2:3153–3165.

Diril et al. PNAS | March 6, 2012 | vol. 109 | no. 10 | 3831 Downloaded by guest on September 27, 2021